Reporter: Aviva Lev-Ari, PhD, RN
Some assays that gauge cancer-related signatures can’t factor in tissue architecture, while other assessments that are good at gauging tissue architecture, provide mostly qualitative tumor data. To reconcile these differences, researchers led by Yinyin Yuan of Cancer Research UK decided to combine histopathological and gene expression analysis to show that quantitative image analysis of the cellular environment inside tumors can bolster the ability of genomic profiling to predict survival in breast cancer patients. This approach, too, though, has its limitations.
For instance, molecular assays that gauge cancer-related signatures are challenged by their inability to factor in tissue architecture and the results are confounded by genomic information from the different types of cells inside the tumor other than cancer cells. Meanwhile, traditional histopathological assessments are good at gauging tissue architecture and differentiating cellular heterogeneity, but mostly provide qualitative tumor data and are too time consuming to be applied in large-scale studies.
Recognizing these weaknesses, researchers led by Yinyin Yuan of Cancer Research UK decided to combine histopathological and gene expression analysis to show that quantitative image analysis of the cellular environment inside tumors can bolster the ability of genomic profiling to predict survival in breast cancer patients. “All technologies have some sort of weakness. That’s why when we combined two types of assays — image and microarray — we get a more reliable readout,” Yuan says.
As they report in Science Translational Medicine, Yuan and her colleagues gathered histopathological information from hematoxylin and eosin-stained images as well as gene expression and copy-number variation data on a discovery set of 323 samples and on a validation set of 241 samples from patients with estrogen receptor-negative breast cancer. Using the discovery sample set, the investigators developed an image-processing method to differentiate the cells inside tumor samples as cancerous, lymphocytic, or stromal. They then tested this technique on the validation sample.
Once Yuan and colleagues had an accurate picture of the types of cells in the tumor samples, they used image analysis to correct copy-number data — as it is influenced by cellular heterogeneity — and developed an algorithm to determine patients’ HER2 status better than copy-number analysis can.
Using the image-processing method, the researchers stratified the discovery and validation sample sets into lymphocytic infiltration-high and lymphocytic infiltration-low groups — as past studies have suggested that high lymphocytic infiltration is linked to better patient outcomes.
When the image analysis was compared to the pathological scores of the samples, the discovery set showed no difference in patient outcomes, but the assessments disagreed with regard to the outcomes of the lymphocytic infiltration-low group in the validation cohort.
Hypothesizing that integrating the gene expression signatures and quantitative image analysis would improve survival prediction, the study investigators combined them. “The gene expression classifier had 67 percent cross-validation accuracy in predicting disease-specific deaths, the image-based classifier had 75 percent, and the integrated classifier reached 86 percent,” the study authors write.
Finally, Yuan and her colleagues applied the image analysis to develop a quantitative score that determines whether specific types of cells are tightly clustered — a high score — or are randomly scattered — a low score. In stromal cells, this approach could discern that breast cancer patients with a high or low score had a “significantly better outcome” than patients whose scores fell in the medium range.
Ultimately, Yuan and her colleagues show that their image processing avoids the biases of manual pathological assessments and accurately quantifies cellular composition and tissue architecture not accounted for by molecular tests. The researchers’ computational approach is also faster than traditional pathological techniques. “These two sets of samples can be done in a day,” Yuan says.
According to the study authors, the limitation of the image processing technique is, of course, that it requires matched molecular and image data.
 |
Turna Ray is the editor of GenomeWeb’s Pharmacogenomics Reporter. She covers pharmacogenomics, personalized medicine, and companion diagnostics. E-mail her here or follow her GenomeWeb Twitter account at @PGxReporter. |
Related Stories
- All the Heterogeneity
November 1, 2012 / Genome Technology
- Decentralized Testing
October 1, 2012 / Genome Technology
- Wrestling Resistance
September 3, 2012 / Genome Technology
- Recruits for Research
April 1, 2012 / Genome Technology
- Study Suggests Flaws in Standard Cognitive Alzheimer’s Test Could Undermine Drug, Biomarker Research
December 21, 2012 / ProteoMonitor
2012pharmaceutical
Aashir Awan, Phd
Alan F. Kaul, PharmD., MS, MBA, FCCP
alexcrystal6
anamikasarkar
apreconasia
aviralvatsa
David Orchard-Webb, PhD
danutdaagmailcom
Demet Sag, Ph.D., CRA, GCP
Daniel Menzin
Dror Nir
evelinacohn
Gail S Thornton
Irina Robu
jdpmd
jshertok
kellyperlman
Ed Kislauskis
larryhbern
lmulligan13gmailcom
marzankhan
megbaker58
ofermar2020
pkandala
Rosalind Codrington, PhD
ritusaxena
Rick Mandahl
sjwilliamspa
stuartlpbi
Dr. Sudipta Saha
tildabarliya
zraviv06
zs22
Posted by Jennifer D. Davis, Ph.D. on LinkedIn, System Biology Group
This is a GREAT approach! I’m happy to see someone is looking into correlative trans-disciplinary approaches. One of the problems with breast cancers is that making a call on the treatment category, for example: ER+ vs. ER- is based on the % of ER+ cells in the tumor — meaning not all of the cancer cells will be ER+. Have you seen the recent nature papers? The Cancer Genome Atlas just published their molecular portrait of human breast tumors and another group the mutational landscape and translocations…its an exciting time for personalized medicine. Now with computational biology to put together predictive algorithms, perhaps one day we will be able to best tailor treatments for patients and improve cure/management rates.
http://www.ncbi.nlm.nih.gov/pubmed/23000897 (TCGA results)
http://www.ncbi.nlm.nih.gov/pubmed/22722202 (analysis of the mutational landscape and translocations)
Posted by Jennifer D. Davis, Ph.D.
This is a big step. Jon Morrow at Yale has to be chomping at the bit to jump on this.
MM Pinto has also gone to more than 1 stain done in sequence, mainly to avoid bringing the patient back for another bite.
I think that it will be another breakthrough to have the laboratory next to the OR, and the pathologist doing this instead of the frozen section.
I thought it was in this post that I saw it, but the frozen section is “riding into the sunset”. The calculations have to be made faster.
Dr. Larry please take upon yourself as your forthcoming contributions for the Cancer book as EAW not Editor the following TWO
http://www.ncbi.nlm.nih.gov/pubmed/23000897 (TCGA results)
http://www.ncbi.nlm.nih.gov/pubmed/22722202 (analysis of the mutational landscape and translocations)
THANK YOU for your leadership in any domain.
If you are to commit to the two above, notify Dr. Williams about it.
No doubt Jennifer,
Several initiatives like the one you mentioned along with other major international efforts in sequencing and mapping a growing number of cancer molecular phenotypes opens up unprecedented opportunities for both scientists and clinicians.
We must consider that genomic analysis will take only so much into the drug sensitivity panels per se compared to transcriptomic and proteomic analysis due to the fact that purely genomic abnormalities take up an average of 5% of all molecular cancer defects but the recently published article you quoted is a good example on how we can use such type of information.
It might take still a few years before we are able to consistently obtain and integrate genomic, transcriptomic and proteomic data from the same patient at an affordable cost but the current speed of innovation observed in multidisciplinary fields seems to point at a switch from evidence based medicine to personalized medicine sooner than previously anticipated.
Dr. Scalia,
Many thanks for your reply, which I happen to be on the same page with you on.
This is very good Jennifer. I have sat through more than enough lectures on people trying to only use a genomics or transcriptomics approach to either classifying tumors or predicting resistance or prognosis and hear that the end result was that one method is never enough. I am glad someone is putting the pathology evaluation back in and would help with some of the concundrums of the mixed pathologies one sees in certain solid tumors as well as hard to detect triple negative breast tumors. I find myself constantly begging the epigenetics and transcriptomics people to do some laser capture microdissection on the tumors (and yes you can get quality RNA and DNA out of parrafin embedded). However I am wondering how this method matches up to immunohistochemical methods people are using such as looking for biomarkers related to chemoresistance in combination with the microscopy and pathologic evaluation.
[…] to predict Survival: Combination of Histopathology and Gene Expression Analysis A Lev-Ari Breast Cancer: Genomic profiling to predict Survival: Combination of Histopathology and Gene Express… (pharmaceuticalintelligence.com) Gastric Cancer: Whole-genome reconstruction and mutational […]
[…] The Initiation and Growth of Molecular Biology and Genomics, Part I […]
[…] to predict Survival: Combination of Histopathology and Gene Expression Analysis A Lev-Ari https://pharmaceuticalintelligence.com/2012/12/24/breast-cancer-genomic-profiling-to-predict-survival-combin… Gastric Cancer: Whole-genome reconstruction and mutational signatures A Lev-Ari […]
[…] Breast Cancer: Genomic profiling to predict Survival: Combination of Histopathology and Gene Express… […]
[…] Breast Cancer: Genomic profiling to predict Survival: Combination of Histopathology and Gene Expression Analysis A Lev-Ari https://pharmaceuticalintelligence.com/2012/12/24/breast-cancer-genomic-profiling-to-predict-survival-combin… […]
[…] Breast Cancer: Genomic profiling to predict Survival: Combination of Histopathology and Gene Expression Analysis A Lev-Ari https://pharmaceuticalintelligence.com/2012/12/24/breast-cancer-genomic-profiling-to-predict-survival… […]
[…] https://pharmaceuticalintelligence.com/2012/12/24/breast-cancer-genomic-profiling-to-predict-survival-combin… […]
[…] Breast Cancer: Genomic profiling to predict Survival: Combination of Histopathology and Gene Express… […]
[…] https://pharmaceuticalintelligence.com/2012/12/24/breast-cancer-genomic-profiling-to-predict-survival… […]